Serchilo:Feature requests
[edit] Default command
After all these months of using Serchilo enthusiastically, I still end up forgetting the little g several times a week when I want to search Google. I propose a default command users can add to their user pages. Something like this:
<command>
default
title = Google-Suche
url = http://www.google.com/search?hl=de&safe=off&q={%query}&btnG=Suche&lr=
</command>
Kevang 21:30, 13 November 2007 (CET)
- I am a bit confused. How does this solve your problem forgetting the "g" command? Arnomane 02:50, 14 November 2007 (CET)
- By not taking me to the serchilo "command not found" page, but directly to the Google search when I forget it. --Kevang 17:48, 14 November 2007 (CET)
- I'm thinking about defining Google as a default command for everybody, if no command is recognized. What do you think about this? --Jorges 03:29, 17 November 2007 (CET)
- I wouldn't recommed that. It may confuse people if they mistype a command and have to go back to serchilo hp. I would make it an optional setting for every user. Lengemann 00:37, 18 November 2007 (CET)
- I agree. --Kevang 13:34, 19 November 2007 (CET)
[edit] A link on the homepage taking you directly to the login form
Would be nice as well. :-)
Kevang 17:48, 14 November 2007 (CET)
- Special:Userlogin is in the upper right corner. Arnomane 19:26, 14 November 2007 (CET)
- I'm talking about the Serchilo Homepage. Kevang 23:48, 14 November 2007 (CET)
- I just put a login form into the footer. --Jorges 03:29, 17 November 2007 (CET)
- Yay, thanks! --Kevang 13:35, 19 November 2007 (CET)
[edit] Command templates
If template parameters in triple braces could be resolved inside <command> blocks, there would be an interesting use for this. Kevang 18:21, 30 November 2007 (CET)
- Sorry for the late answer. You can already use {{subst:MediaWiki}} with parameters to generate the code from the template once. But I don't think it's good idea to keep the template in the certain command page. It would make the command definition very intransparent. Templates would make sense if the target URLs to mediawiki sites would change often, but this doesn't happen. But we may have templates and use them with {{subst: on creating commands. --Jorges
[edit] Multiple Querys with one search
Is there a way for searching all "hello world" videos for example in google and youtube with one query? The query could look like that "g+yt hello world" maybe? As result I would open two tabs or maybe a frame above (like google translate), with links to the other querys/pages.
- The problem is that opening several tabs is done by the client (browser) and depends on what browser is used and how it is configured. I could generate a website that opens via Javascript some new windows, but I am not convinced to this solution yet. Jorges 03:13, 14 January 2008 (CET)
- The variant with using frames could be done with a simple html frameset. But I don't think that's a good idea. It would make the command syntax more complicated and may collide with regexp commands. --Stefan
[edit] Use one Iphone/Ipod Touch
First of all, this is a very useful page for mobile devices. The only real problem is the following: While typing a command, the popup with the hints/helps appears and moves over the field where you type, so you have to blindtype which is not very convienient on a touchscreen.
- As a first workaround I made the suggestions (popup hints) optional. To switch them off, call
http://serchilo.net/?suggestions=off
There will be a mobile interface soon. --Jorges 03:13, 14 January 2008 (CET)- I think it should be possible to adjust the dropdown-CSS/HTML in a way that it will work on several browsers/devices. --Stefan
- This bug reveals a mobile interface apparently exists now. It's great, I think it just needs a viewport metatag to be useful on an iPhone. See http://www.html-metatags.de/351-meta-viewport-mobile-seo/ --Kevang 21:15, 4 October 2010 (UTC)
[edit] Robots-Meta on User Pages
This is not a real Feature Request, but I still feel this should go here.
I just found out that searching my name on Google shows my Serchilo User Page as the 5th element, even before my homepage. Could it be possible to insert a norobots on User Pages? Lengemann 01:07, 24 January 2008 (CET)
- Sorry, I must have overlooked this message. I've put this into the general robots.txt:
User-Agent: * Disallow: /User Disallow: /index.php?title=User
- I hope it solves the problem. --Jorges 01:51, 19 March 2008 (CET)
[edit] HTTP-POST-JavaScript-redirection
I just hacked a small script that generates an HTML post-form out of its GET-attributes and does a submit via Javascript: http://klumpen.ath.cx/labor/postredirector/ (usage example at this page). This should make it possible to access pages with Serchilo that expect their data as POST-data. Maybe someone can try it out with a real life example, I currently just found none. --Stefan
- I used your postredirector with the bible command. Works fine. Thanks for your efforts!
- Great, thanks! I used your script for Altgriechisch-Deutsch. Unfortunately, as with everything, there is a character encoding problem. The function you're using to make the parameter values safe for HTML seems to assume that the input is in some 1-byte encoding (Latin1, probably) so it turns UTF-8 encoded strings into gibberish. Could you fix that or send me your script so I can try? Thanks -- Kevang 11:05, 25 August 2009 (UTC)
- Hi, I just made it use UTF-8 as default and added an optional "_charset"-Parameter for specifying an encoding as listed here: http://php.net/manual/en/function.htmlentities.php --Stefan
- Splendid, now my command works like a charm. Thanks a bunch! --Kevang 01:30, 5 December 2009 (UTC)
- Hi, I just made it use UTF-8 as default and added an optional "_charset"-Parameter for specifying an encoding as listed here: http://php.net/manual/en/function.htmlentities.php --Stefan
[edit] Support for JIS
Support for that encoding would be great because it would allow to enter Kana or Kanji in Jeffrey's Japanese <=> English dictionary. See Japanese-English. --Larso 23:43, 27 July 2008 (UTC)
- I'm just trying to find out more about JIS (and PHP). But are you sure that this dictionary uses pure JIS? It seems, there is also an htmlentities() in between. --Jorges 15:05, 28 July 2008 (UTC)
- Well, I just made a query for はな and also saved a text file containing はな using JIS7 encoding. I then opened the text file in hex editor and compared it to the URL-encoded query. It matched. Query was %1B%24B%24O%24J%1B%28B and the text file was in hex: 1b 24 42 24 4f 24 4a 1b 28 42. And it can't be using Shift JIS because when I saved that text file using that encoding, I get 82 cd 82 c8 in hex. Also tried for query 花形役者 and got the same results. By the way, I didn't get it what you meant with that htmlentities() thing. I only know that it's a PHP function. --Larso 18:42, 28 July 2008 (UTC)
- I just added JIS support. There is a new encoding switch jis that I already added to the argument. It seems to work.
- Is it true that you can only translate into one direction (ja->en)? If yes, the keyword should be ja-en (and another one en-ja for the other direction). --Jorges 11:36, 2 August 2008 (UTC)
- Yeah, it's true. I'll change the keyword right now. I'll also make the other direction now that I realised that it doesn't have to be in Japanese because the whole dictionary works in English. Anyway, thanks :) --Larso 20:45, 2 August 2008 (UTC)
- Well, I just made a query for はな and also saved a text file containing はな using JIS7 encoding. I then opened the text file in hex editor and compared it to the URL-encoded query. It matched. Query was %1B%24B%24O%24J%1B%28B and the text file was in hex: 1b 24 42 24 4f 24 4a 1b 28 42. And it can't be using Shift JIS because when I saved that text file using that encoding, I get 82 cd 82 c8 in hex. Also tried for query 花形役者 and got the same results. By the way, I didn't get it what you meant with that htmlentities() thing. I only know that it's a PHP function. --Larso 18:42, 28 July 2008 (UTC)
[edit] Breaking out of one's user namespace
What's the fastest way to break out of one's user namespace in order to use a command that one has overridden? Prefixing the command with default: seems to work, but that's a lot of characters to type. Using one's language code also works, but it's still two superfluous characters and worse, it forces one to think about which language to use even when that's not at all on the mental agenda. I therefore suggest to allow just a colon in front of the keyword as a namespace break-out mechanism. I'd then have e.g. :la animadvertere as a temporary solution for looking up Latin vocabulary until that overloading bug is fixed. ;) --Kevang 00:00, 13 July 2009 (UTC)
- Done. Just needed to add a command to Switch to language or user. What's the "overloading bug"? --Jorges 23:54, 9 August 2009 (UTC)
- Awesome, thanks! Here's the overloading bug. --Kevang 14:17, 30 September 2009 (UTC)
[edit] writing "#"
hey can someone change serchilo, so that every "#" is converted to "%23". Things like "g c# test" or "msdn c# test" doesn't work correctly.
- Hi, for me your example is working (Firefox with Serchilo as quick search engine). Which browser do you use? --Stefan 21:31, 19 March 2010 (UTC)
- Thanks for your respones. You are right. But it does not work as the default search engine in the address bar.--217.7.244.33 18:41, 11 May 2010 (UTC)
[edit] (semi-)automated check of commands
It would be nice to have some automatism to check the functionality of all commands and to spot these that are orphaned. One simple way would be to extract the examples for each command, test them and check if the server responds with 404. Of course this would not find all bad commands, but might help for a large portion. There should be some API or database dump, to get the needed data for this script. A more advanced method would be, to extend the syntax in a way that every example can get a regular expression that can be matched with the page delivered by the example - like in unit tests. For example "en kartoffelsalat" gets as regexp "/potato salad/" and if some day the regexp doesn't matches anymore there is a high probability that it's a borken command. --Stefan 13:17, 1 October 2010 (UTC)
[edit] automated screenshots of examples
For every command one could generate a screenshot of the page that the command's example redirects to. These screenshots could be displayed at the wiki-pages for the commands or at serchilo.net when browsing thru the autocompletion-proposals. Could be done with khtml2png at fixed time intervals (cron) or every time a command was added or modified. --Stefan 13:17, 1 October 2010 (UTC)
[edit] host YUI-library at serchilo.net
The YUI libraries at the start page that are currently included from yahooapis.com should be hosted at Serchilo's server itself. This is neccessary because currently Yahoo might collect user data (usual server logfiles) which doesn't comply with Serchilo's privacy policy. If Yahoo would be really bad they could collect data the user entered into the autocompletion form. --Stefan 13:17, 1 October 2010 (UTC)
[edit] searching for a command's wiki page is tedious
I often want to directly access a command's Serchilo wiki page, e.g. because its URL needs to be fixed. However, the wiki's search function is not working very well. Try this:
- I want to edit the wiki page for the "pgp" command (which looks up PGP keys on a server)
- I search the wiki for "pgp" : no results (this was before I altered the wiki page)
- I search the wiki for "keyserver": no results
- I search the wiki for "key": 4 results, but the one I want doesn't show up
- I list all commands. It needs 3 more clicks and some scrolling before finding out the wiki page was named "PGP Keysearch".
So all in all, it would be nice to type sth like "command pgp" into serchilo which takes you directly to the wiki page.